The goal of this research project is to develop and implement an MRI-based technique that creates accurate, real-time temperature maps of tissue that are vital to thermal therapy procedures. This research is a vital part of making thermal therapies clinically viable. For thermal therapy procedures to be safe, effective and time-efficient, the tissue in the region of interest must be monitored for temperature changes throughout the treatment. The goal of this research project is to develop and implement 3D magnetic resonance imaging (MRI) temperature measurement techniques to create accurate, robust, real-time temperature maps that can be used to monitor tissue heating during thermal therapy treatments. To achieve this goal, the research will focus on three areas: 1) develop improved temperature measurement techniques and demonstrate they are capable of measuring temperature distributions in tissue; 2) reduce temperature measurement errors that are due to motion; 3) investigate techniques to increase acquisition speed. 1) Three pulse sequences are currently being investigated as possible options to be used in conjunction with, or as an alternative to, the widely popular, but error prone, proton resonance frequency shift (PRFS) technique. A chemical shift spectroscopy pulse sequence, and two 3D single-shot diffusion weighted pulse sequence are all currently operational. Their imaging parameters will be optimized for temperature measurement and they will be tested in heating experiments on ex vivo and in vivo tissue samples. 2) The self referencing capability of the spectroscopic pulse sequence can potentially be used to reduce errors from motion. In ex vivo and in vivo tissue studies, we will test the ability of the spectroscopic sequence to use the fat signal to reduce both motion induced susceptibility errors and motion induced registration errors. 3) Two techniques will be investigated for improving the temporal resolution of the scans. The use of parallel imaging along with reconstruction algorithms such as GRAPPA can be applied to any pulse sequence to decrease the scan time. A temporally constrained reconstruction algorithm has been shown to be capable of reducing scan time and can also be implemented with any pulse sequence. [unreadable] [unreadable] [unreadable]